Easy: it's for the same reason we cannot measure the potential which appears when copper touches iron, or when salt-water touches zinc. This has nothing to do with diodes, with rectifying junctions. Even with Ge and Si, any connection does itproduces a built-in voltage, and no diode-junction need be present: the same problem appears when anya dirty hunk of p-type semiconductor is placed against an uncleaned n-type, or even touches against any metal contact. The problem involves all contacts between all possible conductors, not just diodes. Touch two metals together and they magically become charged.
However, we cannot use a normal voltmeter to measure this steel/copper potential difference. Unfortunately our meter leads are made of metal! If they're copper, then we can touch theour copper meter-probe against the copper piece, but when we touch ourthe second copper meter probe against the steel piece, it produces a backwards potential which cancels out the potential we wanted to measure. The total voltage seen by the voltmeter, adding up the contributions of all metal junctions around the loop, is isalways zero! (If it weren't, then we'd have a perpetual motion machine.) The voltages are really there, but we can't get at them when using metal meter probes.
OK, back to the diodes. When we connect a p-type and n-type semiconductor together, it always creates one of these built-in voltages of roughly 0.6v, even if no rectifying junction is formed. But then when we next add a copper contactcontacts to the ends of the n- and p-type blocks, we create two new built-in voltages at those contact points. And, these metal-semiconductor voltages will exactly cancel the PN junction's built-in potential, just like with the thermocouples. It's The PN voltage is really in there, but we can't get at it with normal meters, since it's being canceled by the metal-semiconductor contacts.
Obviously the built-in voltage in the PN junction has real-world effects: our diodes turn on at around 0.6Vdc. All PN junctions are automatically biased by this built-in voltage. The When a diode is sitting on a shelf in storage, the voltage-force sweeps carriers away from the junction, turning the diode off. If the weird potential wasn't in there, then diodes would turn on at 0Vdc, and would only turn off if reverse-biased.
Weird trivia: solar cells put out a max voltage which is determined by the built-in junction potential. But in the case of solar cells, the potential only appears on the output wires because the sunlight has shorted-out the PN junction! In other words, the output voltage is coming from the metal contacts on the solar cell, not from the PN junction. The PN junction voltage has been removed,removed; reduced to near zero, because the junction is shorted out by the flood of electron-hole carrier pairs. The missing PN junction voltage allows all the other built-in circuit voltages to pump charges through the loop.
More weird trivia: built-in voltages of metal pairs were detected in the early 1800s with electrostatic meters, and when Volta invented his electric pile, he was convinced that the battery potentials were actually these built-in potentials, rather than created by chemical reactions. In other words, Volta was an early Free Energy nutcase, and was convinced that he'd discovered a perpetual motion machine. (His great secret was to avoid copper/tin/copper/tin, but instead to stack up copper/tin/saltwater/copper/tin/saltwater.) Arguments with colleagues broke out, and eventually the scientific community determined that the battery-voltage came from active charge-pumping action, where chemical energy was being removed from the metals as they corroded. TheThe famous electrostatic-motor clocks of DuLuc and Zamboni wouldn't run forever, but only for a few centuries until their metal layers corroded away. The Oxford Electric Bell will eventually stop.
